基于ANN的油水分离双级水力旋流器的结构性能研究

Performance of the Double-stage Hydrocyclone for Oil Water Separation Based on ANN

根据河南油田目前存在的情况,开发了一种井下油水分离系统的双级水力旋流样机。利用人工神经网络(Artificial Neural Networks,ANN)建立起井下双级水力旋流器的数学模型,通过实验室内柴油以及河南油田实际油样测试,根据前后的分析比较来评估该人工神经网络模型的可靠性和有效性。经过最终实验分析,在井下可以实现两级串联油水分离,当流量控制在6m3/h以上,同时入口油滴粒径大于80μm时,分离器的分离效率最好,可以达到99.5%以上。室内实验分离后水中含油浓度小于50PPm,现场试验小于200PPm,远低于国外室内试验的400PPm指标,测量数据可靠,为其应用于实际生产提供了理论分析基础。

Abstract A new double-stage hydrocyclone for downhole oil-water separation system is developed according to the situation existing in Henan oilfield recently. Artificial neural networks is used to establish the mathematical model of the double-stage hydrocyclone downhole, the diesel is tested in the lab and the actual oil samples in Henan oilfield and assess the reliability and effectiveness of the ANN model by com- paring these data. The final experiments show that we can achieve the double-stage hydrocyclone downhole. When flow is controlled at more than 6m3/h and inlet oil droplet size is larger than 80t^m, the separation efficiency of separator is the best. And the oil concentration in water after separating is less than 50 PPm in laboratory and less than 200 PPm in field test, which are far less than 400 PPm of the index in labora- tory experiment of foreign countries. These data is reliable and provide a theoretical analysis foundation for it applying to the practical produc- tion.